Study on the Equivalence Transformation between Blasting Vibration Velocity and Acceleration

Chong Yu, Jiajun Wu, Haibo Li, Yongan Ma, Changjian Wang2024 · DOI

TL;DR

This paper tackles the challenge of converting between blasting vibration velocity and acceleration measurements, crucial for accurate structural safety assessments. By developing a robust noise reduction and transformation method, the authors provide a way to unify these two key vibration metrics, improving data utilization and enabling more informed blasting designs.

What Problem Does This Paper Address?

Blasting operations generate vibrations that can impact surrounding structures. Evaluating the safety of these vibrations typically relies on two primary physical quantities: velocity and acceleration. However, different types of seismometers measure these distinct quantities, and converting between them can be problematic due to signal noise, non-linearity, and non-stationarity inherent in blast vibrations. This makes it difficult to directly compare or utilize data from different monitoring systems or to switch between velocity-based and acceleration-based safety criteria. The paper aims to bridge this gap by proposing a reliable method for the equivalent transformation of blasting vibration velocity and acceleration.

Methods

The core of this research lies in a multi-stage signal processing approach:

  • Trend Removal: The study first addresses the issue of baseline drift in the raw signals. They employ MATLAB's built-in detrend function to remove the best straight-fit line from the data, which is a standard technique for preparing signals for further analysis.
  • Noise Reduction: Recognizing that blast vibration signals are often contaminated with noise, especially at longer distances, the authors propose a novel combination of Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) and Wavelet Decomposition/Wavelet Packet Decomposition (WD/WDP).
  • CEEMDAN is used to decompose the signal into several intrinsic mode functions (IMFs) and a residual term, effectively separating different frequency components and handling non-stationary signals better than traditional methods like Empirical Mode Decomposition (EMD).
  • WD/WDP is then applied to these IMFs. This technique allows for the selection of appropriate thresholds to filter out noisy IMFs while preserving the essential information from the dominant ones. The paper explores various wavelet basis functions and decomposition levels to find optimal noise reduction strategies for both velocity and acceleration signals.
  • Equivalent Transformation: After noise reduction, the study investigates two primary methods for transformation:
  • Acceleration to Velocity: This is achieved through time-domain integration of the acceleration signal.
  • Velocity to Acceleration: This is accomplished through time-domain differentiation of the velocity signal.

Datasets

The study utilizes data collected from field blasting tests conducted at the Haiyang nuclear power plant (HNPP) in Shandong Province, China. The tests involved two blasting methodologies: pre-splitting and bench blasting. A total of 164 sets of data were analyzed, collected from monitoring points at distances ranging from 10 m to 598 m from the blast source.

Key Insights and Ideas

  • Noise is a Major Hurdle: The research strongly emphasizes that raw blast vibration signals, especially at longer distances where the signal-to-noise ratio (SNR) is low, are heavily affected by noise. This noise significantly impacts the accuracy of direct integration or differentiation for transformation.
  • CEEMDAN + WD/WDP Synergy: The combination of CEEMDAN for signal decomposition and WD/WDP for targeted noise filtering proves to be a powerful approach for cleaning up blast vibration signals, significantly improving the accuracy of subsequent transformations.
  • Distance Matters (A Lot): The ratio of peak particle velocity (PPV) to peak particle acceleration (PPA), denoted as Ra, is found to have a positive power-law relationship with propagation distance. This means that as the vibration travels further, the velocity component tends to become relatively larger compared to the acceleration component.
  • Blasting Method Influences Ra: Pre-splitting blasting results in a lower Ra compared to bench blasting. This is attributed to the different objectives and energy conversion mechanisms of these blasting techniques, with pre-splitting leading to higher frequencies and thus faster attenuation of acceleration relative to velocity.
  • Acceleration Attenuates Faster: The attenuation exponent for acceleration is consistently higher than for velocity, meaning acceleration signals decay more rapidly with distance. This further explains the observed increase in Ra with distance.

Key Results

  • Improved Transformation Accuracy: The proposed CEEMDAN and WD/WDP noise reduction method, followed by integration/differentiation, significantly reduces the relative error of the transformation. For instance, at a distance of 598 m, the relative error was reduced to less than 18%, a substantial improvement from the initial errors that could exceed 50% without noise reduction.
  • Ra as a Predictive Tool: The study establishes power-law equations relating Ra to propagation distance (D) and maximum charge per delay (Q). For example, Ra = 1.08891 * D^0.20061 for the overall dataset. These equations, along with prediction bands, offer a means to estimate the ratio between velocity and acceleration.
  • Blasting Method Specifics:
    • Pre-splitting blasting: Ra = 1.08933 * D^0.16807
    • Bench blasting: Ra = 1.13459 * D^0.20047
  • Attenuation Laws: The study provides specific attenuation equations for PPV and PPA, confirming that acceleration attenuates more with distance than velocity.

Limitations and Future Work

  • Limitations:
    • The adjusted coefficient of determination (COD) for the Ra vs. D relationship is noted as being only 0.33614, indicating that propagation distance alone doesn't fully explain the variation in Ra. This suggests other factors are at play.
  • The accuracy of the transformation is heavily dependent on the authenticity and completeness of the original signals, which can be affected by various errors inherent in seismometer measurements.
  • Future Work:
    • The authors suggest that future research should focus on developing an Ra prediction equation that incorporates more variables, such as topography and geology, to improve prediction accuracy.
  • Further investigation into the optimal selection of wavelet basis functions and decomposition levels for different signal characteristics is implied by the detailed analysis of noise reduction indicators.

Summary updated

BibTeX
@misc{yu2024study,
  author = {Chong Yu and Jiajun Wu and Haibo Li and Yongan Ma and Changjian Wang},
  title = {{Study on the Equivalence Transformation between Blasting Vibration Velocity and Acceleration}},
  year = {2024},
  doi = {10.3390/s24061727},
}
Figure